Corruption in Our Courts: What It Looks Like and Where It ...

STRATOS PAHIS

Corruption in Our Courts: What It Looks Like andWhere It Is Hidden

A B ST R ACT. Recent surveys and events indicate that judicial corruption could be a significantproblem in the United States. This Note builds an economic model of bribery to betterunderstand the incentives behind this pernicious phenomenon. It then compiles a data set ofdiscovered incidents of judicial bribery in the United States to test the effectiveness of our anti-judicial-corruption institutions. This analysis suggests that our institutions are particularlyineffective at preventing and uncovering judicial bribery in civil disputes and traffic hearings.

A U T H O R. Yale Law School, J.D. 2009; Universidad Complutense de Madrid, M.A. 2005;

Dartmouth College, B.A. 2004. I am grateful to Professor Susan Rose-Ackerman for inspiringmy interest in this topic and for her invaluable support and feedback over many drafts of thisNote. I am also indebted to the members of The Yale Law Journal Notes Committee-inparticular to Victoria Weatherford for her patience, meticulous editing, and insightfulcomments. Thanks also to Sam Ferguson for his helpful feedback. All errors that remain are ofcourse mine alone.

1900

NOTE CONTENTS

INTRODUCTION 1903

UNDERSTANDING AND OBSERVING JUDICIAL CORRUPTION 19o6

A. An Economic Model of Bribery 19o6

1. Understanding Defendants' and Litigants' Demand for Corruption 19072. Understanding the Supply of Corruption 19o8

B. Morality and the Risk of First Movers 19101. First-Mover Risk 19102. First-Mover Risk with Relaxed Assumptions 1912

C. Model Predictions and Observations 1914

II. SAMPLE AND METHODOLOGY 1916

A. Accountability Institutions 1916B. Sample of Judges and Bribes 1918

C. Methodology 1919D. Sample Bias 1921

III. CORRUPT JUDGES AND COURTS 1922

A. Federal Versus State Judges 1923B. Trial Versus Appellate Judges 1923

C. Elected Versus Appointed Judges 1924D. Multijudge Corruption Rings 1924

IV. FACT PATTERNS OF CORRUPTION 1925

A. Types of Cases 1925

B. Corrupt Actions: What the Bribes Bought 1926

i. Criminal Cases 19262. Civil Cases 19273. Traffic Violations 19274. Administration of Cases 1928

C. Prices: How Much Was Paid 1928

D. Risk of Detection and Tip-Offs 1929

1. Prosecutorial Leverage 19292. Judicial Conduct Organizations and Uninterested Tip-Offs 1930

1901

THE YALE LAW JOURNAL

3. Judicial Extortion 19304. Multijudge Corruption Rings 19315. First-Mover Risk 1931

V. ANALYSIS 1932

A. Criminal Versus Civil Cases 1933

1. Risk of Detection 19332. Actual Incidence 19363. Bias and Alternative Explanations 1940

B. Traffic Bribes 1942

CONCLUSION 1943

1902

118:19oo 2009

CORRUPTION IN OUR COURTS

"If experience demands a presumption that a judge will seize everyopportunity presented to him in the course of his official conduct to line hispockets, no canon of ethics or statute regarding disqualification can save ourjudicial system."

-Justice William Rehnquist'

INTRODUCTION

A judiciary without honesty has little chance of executing its moral andconstitutional duties, no matter how many rules of ethics exist. This isespecially true in the United States, where the judiciary is afforded widediscretion. Facts and law require interpretation; justice and equity requirejudgment. Every decision to grant a motion, to follow precedent, to interpret astatute or facts, to set a sentence or damages -every decision left up to thediscretion of a judge- is a potential opportunity for corruption. Eliminating allopportunities for personal gain would require nothing less than the destructionof the independent and adaptable judicial system we know. And so we counton honest judges to navigate our ship of justice through these dangerouswaters.

But we do not just keep our fingers crossed and hope we have goodcaptains at the helm. We develop processes of choosing the most skilled andhonest judges. We provide them with good pay and professional prestige tolessen the temptations of bribery. And we develop multilevel methods ofoversight that intrude minimally (one hopes) upon their discretion andindependence. We expect judges to be honest because we establish institutionsthat incentivize honesty.

Despite the critical importance of maintaining judicial integrity, there is adearth of empirical literature that analyzes the effectiveness of theseinstitutions. To be sure, some studies have tracked the historical developmentof judicial integrity institutions and others have catalogued cases of judicialcorruption.2 Others still have relied upon questionnaires to gauge perceptions

1. William H. Rehnquist, Sense and Nonsense About Judicial Ethics, 28 REc. ASS'N B. CITY N.Y.694, 699-700 (1973).

2. Books have been written in the past on corrupt judges, but they are historically rather thananalytically focused. See, e.g., CHARLES R. ASHMAN, THE FINEST JUDGES MONEY CAN BUY:AND OTHER FORMS OF JUDICIAL POLLUTION (1973); JOSEPH BORKIN, THE CORRUPT JUDGE:

AN INQUIRY INTO BRIBERY AND OTHER HIGH CRIMES AND MISDEMEANORS IN THE FEDERAL

COURTS (1962). Other works have provided anecdotal evidence of corruption and offeredvarious policy proposals to combat it. See MAx BOOT, OUT OF ORDER: ARROGANCE,

CORRUPTION, AND INCOMPETENCE ON THE BENCH 191-95 (1998) (discussing various cases of

1903


and incidences of corruption.3 But because no study has ventured beyond thedescription of discovered cases of judicial corruption, none has been able toanswer the question of how effective our institutions have been at actuallyunearthing and punishing the crime.4

This Note begins to fill in this serious gap in the literature on judicialcorruption. By developing an economic model to understand judicialcorruption and creating the only recent sample of discovered cases of judicialbribery against which to test its predictions, this Note attempts to assess theeffectiveness of our anticorruption mechanisms. In doing so, beyondcataloguing important patterns in judicial corruption, it advances the argumentthat there is a serious blind spot in the functioning of our anticorruptioninstitutions. While the small sample size limits the certainty of this Note'sfindings, its analysis suggests that the mechanisms for detecting bribery ofjudges in civil matters and traffic violations are deficient and that much judicialcorruption in these cases likely goes unnoticed.

Before moving on, it is worth mentioning why I have specifically focusedthis Note on judicial bribery. After all, many forms of judicial corruption existand may in fact be more widespread than quid pro quo bribery. Cases ofjudgesruling on matters involving a financial or personal conflict of interest arenumerous and are responsible for a large portion of sanctions handed down bystate judicial conduct organizations (JCOs). s The receiving of gifts, thegranting of favors, ex parte communications, and other actions that createpartiality or its appearance are also highly prevalent forms of malfeasance dealtwith by JCOs.

Despite the importance of these forms of corruption, I have chosen to limitmy study to bribery cases for three reasons. First, a recent survey suggests thatjudicial bribery may be a significant problem in the United States. In aTransparency International survey, 2% of the North Americans (defined toinclude residents of the United States and Canada) who had come into contact

judicial corruption); Geoffrey P. Miller, Bad Judges, 83 TEX. L. REV. 431 (2004) (cataloguingcases of judicial malfeasance).

3. See, e.g., Transparency Int'l, How Prevalent Is Bribery in the Judicial Sector?, in GLOBAL

CORRUPTION REPORT: CORRUPTION IN JUDICIAL SYSTEMS 11, 11-12 (Transparency Int'l ed.,2007) (comparing surveys of perceptions of judicial corruption with a poll of the percentageof people who say they have paid bribes within the judicial system).

4. Empirical studies exist on general judicial misconduct handled by judicial conductorganizations - which can include cases of corruption - but they do not make an attempt toanalyze the effectiveness of institutions in dealing with corruption in particular. See, e.g.,CYNTHIA GRAY, A STUDY OF STATE JUDICIAL DISCIPLINE SANCTIONS (2002).

5. See infra Section II.A.

1904

118:19OO 200 9


with the judiciary over the previous year reported having paid bribes.6

Assuming parity of corruption between the United States and Canada, 7 and aU.S. adult population of 220 million,8 this study implies that over one millionbribes are paid in the U.S. judicial system each year. While this survey capturesbribes directed not only toward judges, but also toward police, prosecutors,and jurors, the results are alarming enough to warrant further study intojudges, whose integrity is most critical to a functioning judicial system.Second, cases of bribery offer greater details for study than do other forms ofcorruption. Because bribery is prosecutable, incidents of it should be relativelywell investigated and reported. Third, bribery is one of the most perniciousforms of corruption. It can purchase favors in high-stakes cases and does notnecessitate any personal or professional relationship between the briber andjudge. It would seem, therefore, to be one of the most serious- and difficult todetect-forms of judicial corruption that exists. The most recent judicialscandal to come out of Pennsylvania, in which two judges pled guilty toaccepting bribes from a private juvenile detention facility in exchange forincarcerating minors for extended periods of time, is evidence of just how vileand pernicious the consequences of judicial bribery can be. During the last fiveyears, the judges collected over $2.6 million in bribes and presided over thetrials of five thousand children, including one teenager who was sentenced tofive months detention for stealing DVDs from Walmart.9

This Note is organized as follows: In Part I, I develop an economic modelfor understanding judicial bribery. In Part II, I review the accountabilityinstitutions of the state and federal judiciaries and describe the sample set ofcorrupt judges. I then go over the characteristics of the judges and courts inwhich bribery was discovered in Part III. In Part IV, I discuss the types ofbribery discovered, the prices of the bribes and the corrupt actions that theybought, how the judges and bribers transacted the bribes, and what factors ledto the bribes' discovery. This analysis leads to the troubling observation that

6. Transparency Int'l, supra note 3, at ii.7. It is seems unlikely that judicial bribery in the United States (which ranks twentieth in

Transparency International's (TI) global survey of perceptions of overall corruption) issignificantly less prevalent than in Canada (which ranks ninth in TI's global survey).Transparency International, Corruption Perceptions Index 2007,http ://www.transparency.org/policy-research/surveys_indices/cpi/2o07.

8. The 2004 Census Bureau estimated the adult population to be 217 million in 2004. PressRelease, U.S. Census Bureau, Census Bureau Estimates Number of Adults, Older Peopleand School-Aged Children in States (Mar. 10, 2004), available athttp://www.census.gov/Press-Release/www/releases/archives/population/oo1703.htm-l.

9. Stephanie Chen, Pennsylvania Rocked by 'Jailing Kids for Cash' Scandal, CNN.coM, Feb. 24,2009, http ://www.cnn.conl/2oo9/CRIME/o2/23/pennsylvania.corrupt.judges.

1905


the majority of judges had accepted multiple bribes before being caught andthat some were corrupted by as little as a pound of lunch meat.

In Part V, I examine interesting patterns from the data -in particular, thedisproportionate amount of discovered bribery in criminal cases as comparedto bribery in civil cases. I observe that this discrepancy appears to be due inlarge part to prosecutorial leverage, which allows criminals to bargain downtheir sentence in return for incriminating information about judges, leading toan increased rate of detection. After examining other possible explanations forthe discrepancy, I argue that the data and model support the conclusion thatbribery in civil cases is less likely to be detected than bribery in criminal cases. Iconclude with a summary of my findings and suggestions for further research.

I. UNDERSTANDING AND OBSERVING JUDICIAL CORRUPTION

The study of corruption poses unique problems. Corruption's covert naturemeans that only a fraction of it is ever exposed. Those cases that are discoveredalmost certainly share characteristics that led to their discovery. Relying solelyon discovered cases of corruption as a means of analysis is therefore a limitedmethod that can provide a distorted view of how much and what kind ofcorruption actually exists. This limitation has led scholars to rely on surveydata of public perceptions of corruption as a proxy for the amount ofcorruption that exists.' ° The accuracy or inaccuracy of such perceptionsnotwithstanding, relying solely on public perceptions of corruption is bound toconstrain the specificity of the conclusions. This Part provides anotherframework for understanding judicial corruption.

A. An Economic Model ofBribery

Judicial corruption can be understood as the selling and purchasing of legaldecisions. Understanding judicial bribery requires understanding theincentives that exist for parties or lawyers to purchase these decisions and forjudges to sell them." Below, in an attempt to predict what types of cases and

lo. See, e.g., Daniel Kaufmann, Aart Kraay & Massimo Mastruzzi, Measuring Governance UsingCross-Country Perceptions Data, in INTERNATIONAL HANDBOOK ON THE ECONOMICS OFCORRUPTION 52 (Susan Rose-Ackerman ed., 2006); Paolo Mauro, Corruption and Growth,11o QJ. ECON. 681 (1995); Daniel Treisman, The Causes of Corruption: A Cross-NationalStudy, 76 J. PUB. ECON. 399 (2000).

11. Since Susan Rose-Ackerman's 1978 breakthrough book, Corruption: A Study in PoliticalEconomy, economic analysis has been used widely to understand the interactions betweeninstitutions and corruption. These models primarily focus on the incentives that affect

19o6

118:19oo 2009


judges will yield the most corruption, I describe the major factors thatinfluence the briber's demand for corrupt judicial services and the judge'ssupply of those services. 2

The model, for purposes of simplification, makes two assumptions. First,the model assumes that judges, litigants, and defendants are amoral. Whenthey refuse to engage in corruption it is not because of any moral aversion, butbecause the costs imposed on them for doing so are greater than the gains. Iwill later relax this assumption to help explain the risk involved in engaging incorruption. Second, the model assumes that judges will supply corruptdecisions that, absent corruption, would not be prima facie wrong or inviolation of another law. This assumption is reasonable given the widediscretion judges possess and their desire to limit their exposure to risk ofpunishment, and it is generally borne out in our sample.

i. Understanding Defendants' and Litigants' Demand for Corruption

A litigant or defendant (party) will make the bribe if the expected gains ofthe corrupt decision are greater than the sum of the costs of the bribe and theexpected costs of getting caught. The expected gains (Y) from a corruptdecision can be understood as a function of four factors.

The first and most obvious factor that a party will take into account is thestakes of the case (S). The greater the value of a favorable decision, the more aparty should be willing to pay for it. Assuming that the party had an initialprobability of winning the case (p), the value that the bribe will purchase willbe the added probability of winning the case. In other words, it will be thestakes (S) multiplied by the difference between the original odds of winning(p) and the odds of winning after paying the bribe, which is assumed to be 1, orcertain victory, for the sake of simplicity. So far, the benefit of paying a bribe isequal to (S)(1-p).

individual or group decisions to engage in corrupt behavior, such as the opportunity forcorrupt action, the probability of conviction, the severity of punishment, and theopportunity costs. See, e.g., SUSAN ROSE-ACKERMAN, CORRUPTION AND GOVERNMENT:CAUSES, CONSEQUENCES, AND REFORM (1999); John Macrae, Underdevelopment and theEconomics of Corruption: A Game Theory Approach, 10 WORLD DEV. 677 (1982); AndreiShleifer & Robert W. Vishny, Corruption, io8 QJ. ECON. 599 (1993).

n. The model developed in this Section loosely follows the economic theory of criminalbehavior developed in Gary S. Becker, Crime and Punishment: An Economic Approach, 76 J.POL. ECON. 169 (1968). The discussion of bargaining space loosely follows modelsdeveloped to describe bilateral bargaining and case settlement. See, e.g., George L. Priest,Regulating the Content and Volume of Litigation: An Economic Analysis, 1 SuP. CT. ECON. REV.163 (1982) (developing an economic model for settlement of civil disputes).

1907


We must now take into account the probability that the briber may not beable to keep the spoils of his corrupt decision. This may be the result of twoevents. First, the corrupt decision may be reversed on appeal. The value of acorrupt decision will depend upon how large this rate of reversal (RR) is-aprobability between o and 1. The closer the rate of reversal is to 1, the smallerthe value of the decision. We show this by multiplying our benefit equation by(1-RR). Second, assuming that a party will not be able to keep the spoils of acorrupt decision if the bribe is discovered, the expected gains of the decision ismultiplied by the probability that the corruption will go undiscovered. We canrepresent this by multiplying the equation by (1-r), where r equals theprobability of being caught. The first half of our benefit equation is as follows:

Y= S(i-p)(i-RR)(i-r)

We must now consider the expected losses from offering the bribe. Theseare made up of the cost of the bribe (B) and expected cost of being caught andconvicted (costs of detection or CD). Costs of detection can be considered interms of reputation costs, loss of utility due to time in prison, and opportunitycosts the briber will be forced to pay. To calculate expected losses, we multiplyCD by the risk of being caught (r). Incorporating the price of the bribe, we canrepresent the expected costs of paying a bribe as (r)(CDv) + B. The expectednet benefit, therefore, that a party will experience from "buying" a corruptdecision can be understood in the following terms:

NBp = Y- [(r)(CDp) + B]

The party should be willing to make a bribe as long as it will result in apositive net benefit. In cases, therefore, where the expected gains from acorrupt decision would outstrip the expected losses of being caught, the partywill be willing to make a bribe offer up to a value just below the differencebetween the two. Or, in terms of our equation:

B < Y- (r)(CDp)

2. Understanding the Supply of Corruption

In deciding whether to sell a corrupt decision, a judge is likely to face adifferent decision-making model. The judge does not have to consider thevalue of the decision, the original probability of winning, or the risk of reversal.Instead, the judge will focus on whether the price of the bribe is greater than

1g9o8

118:1 goo 2009


the expected costs of accepting the bribe. As with the briber, the judge willcalculate expected loss by multiplying the probability of being caught (r) by thecost of detection (CDj). The judge's costs of detection are not necessarilyequivalent to those that the briber faces. On average, judges probably facegreater costs of detection in terms of loss of reputation, lost salary, and time inprison. Their detection costs are also relatively higher as judges represent largerprizes for law enforcement who may grant immunity to the briber in exchangefor incriminating information about the judge. The judge is expected to facethe following net benefit equation:

NBj = B - (r)(CDj)

He or she will render a corrupt decision when the bribe is greater than theseexpected costs:

B > (r)(CDj)

Therefore, we expect corruption to flourish when there is a bribe (B) suchthat B > (r)(CDj) and B < Y - (r)(CDp). In other words, a bribe will betransacted when the judge's expected losses are less than the briber's expectednet gains, and there is room for mutual gain.

Graphically this can be represented as follows, where everything to the leftof (Bp(m,)) represents a bribe the party would be willing to pay, and everythingto the right of (Bj(min)) represents a bribe the judge would be willing to take.Smaller bribes on the continuum represent larger net gains to the party, whilelarger bribes on the continuum represent greater gains to the judge.

In Figure 1 below, the overlapping area represents this bargaining spacebetween the judge and the party in which we expect the price of the bribe tofall.

Figure 1: Deal

I Judge's Willingness To Sell

Party's Willingness To Pay I

L0 BPs-W

S BRIBE

In the scenario depicted in Figure 1, we would expect a bribe to changehands, as there is room for mutual gain. By varying the values of the factorsthat affect the decision calculus of the judge and litigant or defendant,

1909


however, we can envision various scenarios where the expected net gains forthe bribing party would be less than the expected costs of the judge.

Figure 2: No Deal

I

IIPat' AIngesT a Judge's Willingness To Sell

I I

L0

S BRIBE

B. Morality and the Risk of First Movers

In the above model, the risks faced by the judge and the briber can each bebroken down into two components. The first component is the risk that theother party reports the bribe to the authorities, and the second is the risk thatthe authorities discover the bribe by some other means. For now, let us focuson the former risk. I call this the first-mover risk, as it is borne in large part bythe party making the initial offer.'3

1. First-Mover Risk

In our model of amoral actors, we would expect the decision whether toreport a bribe to be made based on the net-benefit equations developed above.It is obvious enough that when the party on the receiving end of the offer hasthe opportunity to gain, they will accept it, and when they face a net loss fromthe offer, they will reject it. It is less obvious, however, why an amoral actorwould report a corrupt offer rather than simply reject it. Let us explore why anamoral actor might report a corrupt offer and then examine how relaxing ourassumptions of amorality affects the equation.

13. Various scholars have employed game theory to analyze bilateral corrupt relationships. See,e.g., Melanie Manion, Corruption by Design: Bribery in Chinese Enterprise Licensing, 12 J.L.ECON. & ORG. 167 (1996) (creating a model to determine the payoffs of bribing governmentbureaucrats, some of whom are "clean" and others of whom are corrupt). Melanie Manion'smodel does not, however, factor in the risk of approaching a clean bureaucrat who reportsthe bribe. See also Klaus Abbink, Bernd Irlenbusch & Elke Renner, An Experimental BriberyGame, 18 J.L. ECON. & ORG. 428 (2002) (creating an experiment to analyze the developmentof trust between briber and bribees). The first-mover risk assessed by Abbink, Irlenbusch,and Renner is not the risk that a clean official will report the bribe but that they will take themoney and not deliver on their end of the deal.

1910

118:19goo 2009


a. Scenario One: Judge Offers To Sell a Favorable Decision

In this case, the amoral party, if offered a price that would increase his netbenefit, would accept. If offered a higher nonnegotiable price that woulddecrease his net benefit or if he believed that the judge would not followthrough on his end of the bargain, he would reject the bribe. The party wouldonly report the bribe offer to the authorities in two distinct situations. First, hewould report the offer if he believed a rejection of an unappealing offer wouldlead the judge to hand down an unfavorable decision - that is, if the judge wasextorting from the party an unreasonable sum or if the party believed the judgewas taking bribes from the other side as well. Second, the party would reportthe offer if he could parlay the offer into a favorable outcome for himself. Forexample, criminal defendants could trade information on the corrupt judge inexchange for a more lenient sentence in their present case. An astute judgeshould be able to mitigate this second risk altogether by avoiding engagingcorruptly with criminal defendants and dealing only with civil parties orlawyers who will have little to gain by trading information to prosecutors. Heshould be able to mitigate the first risk significantly by offering realistic resultsat a reasonable price, which, given his insider knowledge of the stakesinvolved, should not be difficult. He would not be able to eliminate the risk,however, unless he could reliably signal that he is not accepting money orbeing pressured from the other side. It may be mitigated, though, by grantingpreliminary motions favorable to the bribe payers before the transaction as away to signal loyalty.

b. Scenario Two: Litigant, Defendant, or Lawyer Offers To Buy aFavorable Decision

The amoral judge would follow a similar cost-benefit analysis as thereceiving party in the above scenario, with a minor difference. The judge couldreject offers that provided net losses without having to worry about thepossibility of retaliation for the rejection. This would leave the judge freesimply to reject the offer without reporting. 4 Even if the law required thejudge to report the bribe, the amoral judge would tend against reporting forthree reasons. First, the bribing party has no incentive to report their own bribeoffer, and so the authorities are very unlikely to discover it. Second, the judgewould avoid reporting the bribe so as not to discourage future offers. Third,

14. Threats of violence, or potential threats of violence, might incentivize reporting in somecases. They also may incentivize the acceptance of unwanted offers.


the judge may avoid reporting for fear that the accused party could turn theaccusation on its head, as was contemplated by the prominent trial lawyerDickie Scruggs, who recently pled guilty to bribery charges.'I

Given this analysis, it seems that while neither of the first movers in ouramoral world appear to face significant risk, a judge would face a comparativelygreater risk in initiating the first move.

2. First-Mover Risk with Relaxed Assumptions

Thankfully, the real world provides at least one additional obstacle tocorrupt transactions that was assumed away in the previous Section: morality.We can assume a moral, law-abiding person will face not only a loss of utilityfrom engaging in corruption but also an increase in utility from reportingcorruption and enforcing justice. The first mover, however, is likely unaware ofwhether and by how much these considerations affect the other party's cost-benefit equation. For some moral actors, acting corruptly and not reporting thebribe may present a finite cost that can be compensated by other gains. Forothers, it may represent an infinite cost, making them totally incorruptible. Wemay incorporate the moral cost of making a bribe into our model, on thelitigant's side, by including it as a value (M,), such that when the party makes abribe, the value of M, will be negative. On the other hand, when he reports abribe request by a judge, the value of M, will be o, but he will reap a utility gainof M2 from acting morally. This is represented below, where the first equationrepresents the net benefit equation of a moral party making a bribe, while thesecond equation represents the net benefit of the moral party choosing toreport the bribe.

(1) NB, = Y- [(r)(CDp) + B] -M

(2) NB 2 = M,

The party, therefore, will accept a bribe request from a judge whenY - [(r)(CDp) + B] - M, > M,. In terms of the size of the bribe, the moral partyis willing to make any bribe (B) such that

B < Y- (r)(CDp) -M,-M,

15. Jonathan D. Glater, Guilty Plea by Lawyer to Bribery, N.Y. TIMES, Mar. 15, 2008, at Ci;Nelson D. Schwartz, The Legal Trail in a Delta Drama: Trial ofLeading Lawyer May Hinge onAlly's Role, N.Y. TIMEs, Jan. 20, 2008, Sunday Business, at i (describing that accusing thejudge of soliciting the bribe was Scruggs's prospective defense strategy).

1912

118:19goo 200 9


Any bribe request larger than that will be reported.Graphically, we can illustrate this as follows. If requested by the judge, the

amoral party would be willing to make a bribe of up to a given bribe, B. If thatsame party developed ethical scruples, he would suffer moral costs for payingthe bribe, represented by the lightly shaded region, and would therefore onlybe willing to make a bribe of only (B - M,). It is important to remember that abribe of (B - M) would leave this party as well off as not making a bribe in thefirst place. The gains he experiences increase from this indifference point as thesize of the bribe decreases. Therefore, once the bribe reaches and passes (B -M,), the party will compare the gains he would receive from making the bribewith the opportunity cost of M2 he would incur from not reporting the briberequest, represented by the darkly shaded region. Only once the gains from thecorrupt decision compensate the party for this opportunity cost would he makethe bribe. In all other circumstances, he would report the solicitation by thejudge. The same dynamic would hold if the litigant or the defendant wereacting as the first mover. In the scenario illustrated below, any bribe requestthat is greater than (B - M, - M 2) would be reported.

Figure 3

Moral Party's Willingness To Pay I

Amoral Party's Willingness To Pay I ?

0 -MI-M2 B-Mi B

S BRIBE

Because moral values are not always obvious on the surface, this presents asignificant obstacle to the first mover in the real world. This informationalasymmetry is probably enough to prevent a first bribe between many partiesthat would otherwise both benefit from a transaction. We would expect bribersand judges to attempt to get around this by offering noncommittal signals oftheir willingness to engage. We would also expect first-time corruptrelationships to be formed around relatively high-stakes cases. Corrupt judgesmaking the first move will be able to offer a lot of value to high-stakes parties,increasing the likelihood that their moral losses will be compensated.Conversely, because first-moving parties in high stakes cases have the most togain, they also have the most to give, making it more likely their bribe offercould entice a morally flexible judge.

If and when this first-mover obstacle is overcome and parties areconnected, they are likely to enjoy more open and fruitful negotiation and a

1913


considerable decrease in risk. As a consequence, we might expect corruptjudges to engage in multiple transactions with the same parties. At the sametime, multiple favorable rulings to the benefit of one party or one lawyer mightraise red flags that increase the risk of detection by other means. The extent towhich judges and parties engage in multiple transactions with one another willdepend upon how big this risk is relative to the risk of the first mover.

C. Model Predictions and Observations

Laying out the multiple predictions that could be derived from the variouspossible combinations of factors described above would be distracting andunhelpful. Instead, in the following Parts, I will use the above model as a lensfor observing interesting patterns in the sample, discussing the patterns thatseem to confirm or challenge my assumptions. For the sake of space, I will notnecessarily revert to the full equations, but limit the discussion to only therelevant factors involved.

Because this Note is primarily concerned with the effectiveness of ouranticorruption institutions, however, it is worth discussing how the risk ofdetection incorporated in the above model would affect observations of briberyin the sample. A higher rate of detection in some cases or for some judgesshould lead to a bias in our sample toward those cases. At the same time, weexpect judges to internalize this risk by requiring a larger payment per bribeand expect bribing parties to internalize the risk by offering smaller bribepayments or not offering them at all. All things equal, a higher rate ofdetection, therefore, should reduce the overall incidence of corruption. Wewould not expect, however, for this reduction to necessarily balance out thebias in the sample of observed corruption away from those types of cases. Thefollowing example is illustrative.

Suppose that all judges face the same costs of detection-for example,$10,000. Now, suppose as well that there are two types of cases: one nonrisky,with a 10% chance of detection and another risky, with a 20% risk of detection.Under this scenario, a judge in a nonrisky case would require a bribe that is atleast equal to his expected loss of $1ooo (o.i x $1o,ooo) to act corruptly. Thissame judge would require at least $2000 (0.2 X $10,000) to act corruptly in arisky case.

Now, also suppose that the net benefit of corrupt decisions for ourpotential bribers ranges between $o and $3000, and, to keep things simple,that there is an equivalent uniform distribution of these benefits for both riskyand nonrisky cases. This is depicted graphically below, where net benefit isdepicted on the x-axis and a function of net benefit (,f(B)) represents thenumber of parties at each bribe level on the y-axis. The parties' ranges of net

1914

118: i9oo 2009


benefits are represented by the rectangle. The judges' acceptable bribe rangefor risky cases is represented by the upper arrow, while the lower arrowrepresents their acceptable bribe range for nonrisky cases.

Figure 4

Judges'Wifog .... To Sell (RiskyCases)....

012 34

S Bribe Cin thousands)

As we can see from Figure 4, there will be bribes exchanged in two-thirdsof nonrisky cases, represented by both shaded regions, while only one-third ofrisky cases will be corrupted, represented by the darkly shaded region. Butwhat will be observed is much different. To illustrate, assume there are a totalof thirty risky cases and thirty nonrisky cases. The breakdown from Figure 4shows that in twenty of the nonrisky cases bribes will be exchanged, while onlyten bribes will be transacted in the risky cases. By multiplying the number ofrisky and nonrisky bribes by their respective rates of detection, we note thepublic will discover two instances of each type of case. Therefore, despite theactual difference in incidence of bribes, different rates of detection will lead usto observe the same number of bribes for each type of case. And unless theincidence of risky bribes decreases to zero, the bribes we observe will be ofhigher value than the actual distribution.

This example helps to illustrate the pitfall involved in studying observedcorruption: what you see is almost certainly not what you get. While there isno way to guarantee an accurate analysis, accuracy can be maximized byunderstanding risk and the other incentives that drive parties to act corruptlyor not. The framework built in this Part is the first step in this effort. Thesecond step will be taken in the following sections, which investigate, amongother aspects, the details of how cases of corruption were discovered and bywhat institutions. The next Part begins with a quick review of theseinstitutions before describing the sample of corrupt judges they turned up.


II. SAMPLE AND METHODOLOGY

A. Accountability Institutions

Both state and federal judges face sanctions for corruption from primarilythree institutions16: (1) impeachment by Congress or the state legislature;(2) criminal prosecution by federal or state authorities; and (3) censure by thefederal judicial councils (FJCs) or, in the case of state judges, censure orremoval by state JCOs.

While impeachment by the legislative branch was originally the preferredmethod of accountability for both state and federal judges, this time-consuming and political-capital-draining process 7 has been used lessfrequently as the obligations of Congress and state legislators have grown. 8

The criminal prosecution of judges, which only became an acceptedpractice with regard to Article III judges after 1973," has been filling this void.Tellingly, the last five federal judges to be impeached by the House andconvicted by the Senate already had been either convicted of, or charged with, acrime.2 ° Prior to their convictions, fifty years had passed without a singleimpeachment. Criminal prosecution of judges, while posing potential threatsto judicial independence, is facilitated by prosecutors' capacity to offer pleabargains and immunity to informants. Indeed, most of the tips that theDepartment of Justice (DOJ) receives related to judicial corruption come fromcriminal defendants or convicts." Prosecution can also be effective in breakinginto otherwise secretive relationships, such as multijudge corruption rings.

16. Many states have gone a step further and introduced judicial elections as an accountabilitymechanism. Although analyzing their success in holding corrupt judges accountable mightmake for an interesting study, it would require data and an approach that fall outside of thescope of the present analysis.

17. See Warren S. Grimes, Hundred-Ton-Gun Control: Preserving Impeachment as the ExclusiveRemoval Mechanismfor Federal Judges, 38 UCLA L. REV. 1209,1210 (1991).

18. See id. at 1224.

19. Federal judges had been investigated by the Department of Justice as early as 1913, but untilthe 7th Circuit in United States v. Isaacs, 493 F.2d 1124, 1142-44 (7th Cir. 1974), cert denied subnom Kerner v. U.S., 417 U.S. 976 (1974), ruled that impeachment was not required beforethe prosecution of federal judges, prosecution did not become accepted practice. See generallyEMILY FIELD VAN TASSEL, BEVERLY HUDSON WIRTZ & PETER WONDERS, WHY JUDGESRESIGN: INFLUENCES ON FEDERAL JUDICIAL SERVICE, 1789 TO 1992, at 34-35 (1993)(discussing the history of judicial prosecutions).

2o. See Grimes, supra note 17, at 1214-15.21. See Reid Weingarten, Judicial Misconduct: A View from the Department of Justice, 76 KY. L.J.

799, 8oo (1988).

1916

118:19cOO 2009)


Although lacking the leverage available to prosecutors, federal judicialcouncils,22 and JCOs, which now exist in all fifty states,23 have greatly eclipsedCongress and state legislatures as anticorruption institutions. The former arecomposed of federal district and circuit court judges who investigatecomplaints filed by any citizen, which are vetted by the chief judge of thecircuit. The Act creating the FJCs expressly withholds from them the authorityto remove federal judges. Instead, when the investigatory committeeuncovers criminal or impeachable offenses, the judicial council may refer thejudge to the Judicial Conference of the United States-an umbrellaorganization composed of judges elected from all circuits and presided over bythe Chief Justice of the United States.2" The Judicial Conference may furtherinvestigate the matter and certify the grounds for impeachment to the House.26

State JCOs, which receive and investigate complaints similar to federaljudicial councils, offer a more streamlined approach to removal than theirfederal counterparts. In most states, these organizations are comprised of a mixof lawyers, judges, and lay people27 and use the American Bar Association'sModel Code of Judicial Ethics supplemented with statutory or constitutionalprovisions as their guidelines for determining misconduct.28 They generallyhave the authority to investigate complaints made against judges, to issue aprivate admonition, to publicly reprimand or censure, to suspend, to imposemandatory retirement, to issue fines, and to recommend removal from office.29

Rather than referring egregious conduct to the legislature for impeachment,most states have a process by which the JCO seeks removal by the state'shighest court, or in some cases, another independent committee.30

22. Federal judicial councils were established by the Judicial Councils Reform and JudicialConduct and Disability Act of 198o, Pub. L. No. 96-458, 94 Stat. 2035 (codified at 28 U.S.C.§ 351-364 (2000)).

23. See MARY L. VOLCANSEK, MARIA ELISABETTA DE FRANcIscIs & JACQUELINE LUCIENNE LAFON,

JUDICIAL MISCONDUCT: A CRoss-NATIONAL COMPARISON 1o8 (1996).

24. § (3)(c)(6)(B)(vii), 94 Stat. at 2037 (codified at z8 U.S.C. § 354).

25. SeeVOLCANSEK ET AL., supra note 23, at 102.

26. See Grimes, supra note 17, at 1221.

27. VOLCANSEK ET AL., supra note 23, at 1o8.

28. See James D. Miller, State Disciplinary Proceedings and the Impartiality of Judges, in STATE

JUDICIARIES AND IMPARTIALITY: JUDGING THE JUDGES 119, 119 (Roger Clegg & James D.Miller eds., 1996).

29. See James Alfini, Shailey Gupta-Brietzke & James F. McMartin IV, Dealing with JudicialMisconduct in the States: Judicial Independence, Accountability and Reform, 48 S. TEx. L. REV.889, 892 (2007).

30. See id.


B. Sample ofJudges and Bribes

The sample31 used for this study is comprised of twenty investigations ofjudicial corruption from 1967 to 2000, where a total of thirty-eight state andfederal judges were either convicted or removed from office by the institutionsdescribed above, on charges related to quid pro quo bribery in their traditionalcapacity as judges.32 The cases were gathered from three main sources: theJudicial Discipline and Disability Digests 196o-1991, 33 which summarize alljudicial discipline cases (related to removal or censure, but not criminal cases)handled by JCOs or state courts for those years; Cynthia Gray's A Study of StateJudicial Discipline Sanctions,34 which chronicles every removal of state judges byJCOs or the state courts from 199o to 2ooi; and the DOJ's annual reports toCongress on the activities and operations of the Public Integrity Section from1978 to 2ooo (PIN Reports)." The Public Integrity Section (PIN) prosecutescases that pose a conflict to the U.S. Attorney of the district or that are

31. See Stratos Pahis, Table Accompanying Corruption in Our Courts (June 1, 2009)

(unpublished manuscript), available at http://yalelawjoumal.org/images/pdfs/pahis-table.pdf. Reconstructing the fact patterns contained in the Table required piecingtogether facts from a variety of sources. It was, therefore, unworkable to include pincites foreach fact included. Instead, I aggregated the relevant sources in a distinct footnote for eachindividual judge in the table.

32. I define the traditional role as one related to the administering, hearing, and deciding ofcases. I do not include judges removed or convicted for bribery relating to administrativefunctions, unrelated to court business. See, e.g., United States v. Reed, 647 F.2d 849 (8thCit. 1981) (deciding a case in which an Arkansas judge charged with perjury in connectionwith allegations that he accepted bribes in his capacity as county administrator). I did,however, include three federal judges-Judge Alcee Hastings and Chief Judge WalterNixon, who were both convicted of perjury, and Chief Judge Harry Claiborne, who wasconvicted of tax evasion. All convictions were related to bribery charges and the three judgeswere eventually impeached and convicted by the Senate. Given the significance of the cases,the strength of the government's evidence, and the eventual impeachments, I thought it wasappropriate to include them in the study. I also included Judge Edward DeSaulnier, whowas investigated, censured, and disbarred by the Massachusetts Commission on JudicialConduct, and who was referred to the state legislature for an impeachment investigation butresigned before action was necessary.

33. JUDICIAL DISCIPLINE AND DISABILITY DIGEST: 1989-1991 SUPPLEMENT (Sara Mathias ed.,1993); JUDICIAL DISCIPLINE AND DISABILITY DIGEST: JULY 1986-DECEMBER 1988 SUPPLEMENT

(Sara Mathias, Anne Lawson & Sheila MacManus eds., 1990); JUDICIAL DISCIPLINE AND

DISABILITY DIGEST: POST-1980 SUPPLEMENTS (Yvette Begue ed., 1988); JUDICIAL DISCIPLINE

AND DISABILITY DIGEST: 1960-1978 (Judith Rosenbaum ed., 1981).

34. GRAY, supra note 4.

35. Archive of PIN Reports, http://www.usdoj.gov/criminal/pin (last visited May 10, 2009).

1918

118:19OO 2009


politically sensitive.36 Consequently, the PIN handles all cases involving federaljudicial corruption 7 and at least some cases of state judicial corruption." Ofthe investigations found in these sources, only those for which the fact patternsof corruption were available in major newspapers 39 or in the Westlaw legaldatabase were kept for this study.

This collection of cases represent, to the best of my knowledge, everyconviction or impeachment related to bribery of a U.S. federal judge from 1967

to 2000. The collection also includes most removals of state judges by a JCO orstate court on charges related to bribery, and most bribery-related convictionsof state judges stemming from prosecution by the PIN and reported in the PINReports for those same years. Missing from the study are cases in which statejudges were prosecuted by U.S. Attorneys, rather than the PIN, and judgeswho resigned or died before they could be removed. Also missing are caseswhere state judges were impeached, although this appears to be a minoromission.

40

C. Methodology

The judges in the sample are compiled in a table available as a PDF on TheYale Law Journal website.4" The Table also includes the following information:(1) the date of the conviction or removal from office; (2) the court the judge saton and whether the judge was appointed or elected; (3) the number of bribesthe judge accepted and the types of cases for which they were accepted; (4) thecorrupt action the judge took or promised to take in consideration of the bribe;(5) whether the bribe was transacted directly through the parties, through the

36. U.S. DEP'T OF JUSTICE, REPORT TO CONGRESS ON THE ACTVITIES AND OPERATIONS OF THE

PUBLIC INTEGRITY SECTION FOR 2000, at 2 (2000) [hereinafter PIN REPORT 2000], availableat http://www.usdoj.gov/criminal/pin/docs/arpt-200o.pdf.

37. See Weingarten, supra note 21, at 799.

38. See infra note 44 and accompanying text.

39. Using Proquest, I searched the following newspapers: Atlanta Journal-Constitution, BostonGlobe, Chicago Tribune, Los Angeles Times, New York Times, Wall Street Journal, andWashington Post. When information on an investigation was available in this database or inthe Westlaw legal database but was not adequate for my purposes, I expanded the search toinclude the Pittsburgh Post-Gazette, Philadelphia Inquirer, Philadelphia Daily News, St. LouisPost-Dispatch, and San Jose Mercury News.

40. Impeachments of state judges appear to be a rare occurrence. In the last fifteen years, onlytwo state judges have been impeached, and one convicted. American Judicature Society,Methods of Removing State Judges, http://www.ajs.org/ethics/eth-impeachement.asp (lastvisited May iO, 2009).

41. Pahis, supra note 31.

1919


lawyers, or through a middleman; (6) whether the judge accepted bribes frommultiple parties or lawyers; (7) the size of the bribes; (8) how the corruptionwas uncovered; and (9) what charges the judge faced and, if applicable, how hewas removed. With regards to types of cases in category 3, I create fourclassifications: criminal, civil, traffic, and administration. Civil cases aredefined by two private entities facing off against each other, while criminal andtraffic cases involve private individuals facing off against the government.42 Idefine traffic cases to include regular traffic violations, as well as ordinanceviolations and drunk-driving charges a.4 The classification of administrationdoes not refer to a particular type of case but rather to the judge's managementof courtroom business. Corrupt administrative actions, for example, includesetting bail and assigning lawyers to clients in return for kickbacks.

Throughout the Note, I make repeated references to information containedwithin the Table. When I do so, I cite in footnotes to the last names of specificjudges whose corrupt acts are described therein. Generally, when discussingthe attributes of the type of cases or bribes it is more appropriate to discuss thesample in terms of aggregate number of bribes rather than the judges inparticular. Some vagaries in the fact patterns, however, inhibit the degree towhich counting bribes is possible. In some of these situations, I revert tocounting judges as a proxy for bribes. When analyzing the effectiveness of ouranticorruption institutions, on the other hand, it makes more sense to talk interms of individual judges, the specific bribes that led to their discovery, andthe investigations that ensnared them.

When counting and analyzing bribes, I intend to describe not the numberof transactions, but rather the number of corrupt decisions the bribes weremeant to purchase. This approach is appropriate, as in many instances multiplebribes are made as installments toward the same corrupt action. Aggregatingthem allows us to analyze how much that action was worth. In other instances,a single payment is made to purchase influence in a number of cases. Whenpossible, I will count the number of cases that the briber intended to influence

42. There are civil cases in which the government is a party, but those do not arise in thesample.

43. I include ordinance violations in accordance with the methodology of the COURT STATISTICS

PROJECT, EXAMINING THE WORK OF STATE COURTS 12 (2007), available athttp://www.ncsconline.org/D-Research/csp/2oo7-fies/Examining%2oFinal%20 %202007%2o-%2o1%2o-%2oWhole%2oDoc.pdf. I include drunk-driving cases within the trafficcategory because many of the judges in the sample were caught for accepting bribes in bothregular traffic cases and drunk-driving cases and often the fact patters did not allow foraccurately distinguishing between the two. See infra notes 62, 64 (discussing of how thisdefinition might affect the analysis).

1920

118:1900 2009


as distinct bribes. When these numbers are not available, I will instead countthe number of payments made.

While information regarding most of the judges in the sample is fairlycomprehensive, not all categories of information are available at the same levelof detail for all judges. For example, there are some notable gaps in a numberof the judges who were exposed in an expansive investigation by the PIN,dubbed "Operation Greylord," into corruption within the Cook County Courtof Chicago, Illinois. All told, the investigation convicted fifteen judges (each ofwhom is included in the sample) and fifty lawyers, in addition to police officersand court clerks.4 Because the investigation uncovered corruption that in somecases spanned years, information regarding the number of bribes exchanged issometimes unclear. Many judges were involved in kickback schemes throughwhich they steered unrepresented clients to lawyers and then recouped aportion of the fees paid to the lawyers. Because the practice was so widespreadand continued over such a long period of time, newspaper and court documentaccounts often do not describe in numerical terms the number of bribes thatwere exchanged. In other circumstances, the judges pled guilty before therecould be a public airing of all of the charges, although it is clear they wereinvolved in ongoing corruption. In these instances, I indicate the number ofbribes in the Table as "multiple" and do not include these in the bribe counts,although I will make mention of how they might affect the conclusions.

D. Sample Bias

Unfortunately, because this study does not include every discovered case ofjudicial corruption, there is the possibility that the sample is biased. Asdescribed above, I include only cases of corruption in which judges wereremoved by JCOs or prosecuted by the PIN, and for which details wereavailable in either the Westlaw legal database or in a major newspaper. Thisselection probably favors cases of corruption involving multiple and prestigiousjudges, large stakes, and egregious corruption, since such cases' scope andramifications are more likely to grab the attention of the PIN and the majormedia. This bias is ameliorated by the fact that many of the cases' details camefrom removal hearings. The amount of detail offered in the decisions of theremoval hearings varied considerably, but the level of detail is most likely notinfluenced by any of the above biasing factors. The PIN's exclusive jurisdiction

44. Maurice Possley, August 5, 1983: Operation Greylord Investigation Revealed, CHI. TRIB., Nov. 6,1997, § 2, at 2. For background on Operation Greylord, see JAMES TuoHY & ROB WARDEN,

GREYLORD: JUSTICE, CHICAGO STYLE (1989).


over federal judges, however, probably biases the sample away from statejudges. I will discuss in Parts TV and V how this bias might influence theobservations.

Ill.CORRUPT JUDGES AND COURTS

Currently, of the total number of federal and state judges on the bench,state judges account for more than 31,000 (_95%),41 and federal judges,including bankruptcy and federal magistrate judges, account for the remaining166o (_5%).46 The 38 judges studied yielded a similar proportion, with statejudges accounting for 34 (-89%) of the judges removed, and federal judgesaccounting for 4 (-11%) of the judges removed. Comparing the number ofbribes that were discovered to the caseloads of state and federal judges is moredifficult to do accurately, given the vagaries in the fact patterns. But accountingfor those bribes for which there is a record, the sample shows five bribesaccepted by federal judges were discovered (0.2%) compared to over 2840bribes by state judges (99.8%). This shows that corruption by federal judges isunderrepresented in our sample compared with the proportion of totalincoming trial cases that the federal and state courts respectively accept. In2006, state courts handled approximately 98% of total combined incomingtrial cases, while federal courts handled the remaining 2%. 47

All judges in the study served at the trial level. It is not surprising that casesof appellate corruption are minimal, considering that judges of first instancerepresent 98.3% of the state judiciary48 and 89% of the federal judiciary.'

Of the state judges removed or convicted, 29 of the 34 judges (85%) wereelected. This is almost identical to the 87% of state trial and appellate judges

45. As of 2oo3, there were 1361 state appellate judges and about 30,000 state trial judges. COURTSTATISTICS PROJECT, supra note 43, at 13, 63.

46. There are 179 federal circuit court judges, 28 U.S.C. § 44(a) (2000), and 663 federal districtjudges, id. § 133(a). Additionally, there are 316 bankruptcy judges, id. § 152(a)(2), and about540 federal magistrate judges, Federal Magistrate Judges Association,http://www.fedjudge.org (last visited May 1O, 2009).

47. There were 102.4 million cases filed in state courts in 2006. COURT STATISTICS PROJECT,

supra note 43, at 12. There were about 2.1 million cases filed in federal courts overapproximately the same period (the twelve-month period ending on March 31, 2006).Federal Judicial Caseload Statistics, Judicial Caseload Indicators,http://www.uscourts.gov/caseload2oo7/frontIlndicatorsMaro7.pdf (last visited May to,2009).

48. See supra note 45.

49. See supra note 46.

1922

118:1900 2009


who either gain or retain their posts through elections."0 Counting the numberof bribes accepted by elected versus appointed judges shows a different result.Of the total number of bribes accepted by state judges, only 14 bribes acceptedby appointed state judges were discovered (0.5%), while over 2700 bribes byelected judges were discovered (99.5%).

A. Federal Versus State Judges

Because federal judges hold positions of higher prestige and pay relative tostate judges and therefore face higher detection costs, we expect their supply ofcorruption to be comparatively low. And because federal judges deal with casesof relatively higher stakes, we expect the demand for federal corruption to becomparatively high. According to our model, then, our net expectation is thatbribe prices in the federal judiciary would be comparatively high, while the neteffect on quantity supplied would depend upon which changed more: demandor supply.

While the sample size is too small to draw any significant conclusions, itprovides anecdotal evidence that our expectations are indeed correct. Three ofthe federal judges in the study drew the largest bribes of the study- $15o,ooo,$100,000, and $85,ooo. The smaller incidence of discovered bribery in federalcourts with respect to the number of federal cases may suggest that theincreased detection costs of judges outstrips any increases in the willingness topay of parties before federal courts. This also may be attributable, however, tothe different makeup of cases before federal and state courts.5 '

B. Trial Versus Appellate Judges

The sample seems to support the notion that bribery in courts of appeals isless common than in trial courts. Every one of the judges studied was bribed atthe trial level. As noted, appellate judges represent just a fraction of the stateand federal judiciaries, and so the absence of appellate bribery in the studysimply could be attributable to the small fraction of cases that appellate courtshear. But there is reason to believe that appellate courts are structurally lessprone to bribery. Appellate judges are more carefully vetted (and therefore

50. Robert C. Berness, Norms ofjudicial Behavior: Understanding Restrictions on Judicial CandidateSpeech in the Age ofAttack Politics, 53 RUTGERS L. REV. 1027, 1028 (2001).

51. Incoming federal trial cases are proportioned as follows: 11.5% civil; 3% criminal; 85%bankruptcy. Federal Judicial Caseload Statistics, supra note 47. Incoming state trial cases aredivided as follows: 17% civil, 21% criminal, 54% traffic, 6% domestic, and 2% juvenile.COURT STATISTICS PROJECT, supra note 43, at 12.

1923


perhaps face higher morality costs), receive higher pay, and hold positions ofgreater esteem. Of course, appellate cases are less likely to be reviewed again bystate high courts or the Supreme Court, which makes bribery more attractive atthis level. But the effect of the lower rate of reversal may be mitigated by thefact that appellate judges often must decide cases in panels of three or more.Successfully corrupting a decision, therefore, requires bribing two or morejudges, which raises the price of the bribe and the risk of being caught for boththe bribing party and the judges involved."2

C. Elected Versus Appointed Judges

Given the copious criticism of judicial elections as a poor method ofselecting qualified judges, we might expect elected judges to fare worse in thisstudy than unelected judges. According to this sample, a similar proportion ofelected judges were caught acting corruptly as unelected judges, but electedjudges were caught accepting a larger number of bribes relative to the numberof cases that they handle.13

D. Multijudge Corruption Rings

Finally, it is worth noting that three investigations were responsible forprosecuting twenty-one of the thirty-eight judges in the sample. While itwould seem this sample might be biased toward cases of large-scalecorruption, 4 these cases do seem to intimate that corruption has a potentiallyinfective quality and flourishes when those higher up in the hierarchicalstructure engage in it. In all three cases of large-scale corruption studied here,the supervising judge was corrupt and, in at least two of the cases, he appearedto gain the most from the corruption scheme. In Subsection IV.D.4 , I discussthe risk implications for multijudge bribery schemes.

52. See infra Subsection IV.D.4.

53. This assumes that elected judges handle a similar proportion of cases as appointed judges.

54. See supra Section II.C.

1924

118:19goo 200 9


IV. FACT PATTERNS OF CORRUPTION

A. Types of Cases

This study reveals that the most common types of judicial bribery that arediscovered and punished are bribes related to traffic violations and criminalprosecutions. At least twelve judges accepted a bribe in a traffic-violation,drunk-driving, or ordinance-violation case."5 At least sixteen judges accepted abribe related to a criminal case. s6 Meanwhile, removal or conviction forcorruption in civil cases was far less prevalent; only five judges were disciplinedfor such offenses. 7 In addition, at least eleven judges received kickbacks fromattorneys whom the judges either appointed, or steered unrepresenteddefendants to, or allowed to solicit clients in their court. s8 Another judge wasremoved for receiving kickbacks from a bail-bondsman, 9 while yet anotherwas convicted of accepting bribes in a licensing court.6,

Comparing the actual number of cases in which bribes were discoveredrather than the number of judges who accepted the bribes is in theory a moreaccurate method of analyzing the distribution of bribery across cases. Givensome of the vagaries in the fact patterns, this is difficult to execute in a precisemanner. For example, some judges from Operation Greylord were indicted fortaking an unidentified number of bribes to fix traffic cases. But even countingjust the number of bribes that were identified, there are over 2500 traffic bribeswithin the sample. 6' Given that the sample documents only about one hundrednontraffic bribes that were meant to influence the judge's disposition toward acase, these transactions place traffic cases in a clearly dominant position vis-a-vis criminal and civil cases. They also would seem to make traffic bribery quiteoverrepresented in our sample relative to the proportion of traffic cases

55. Judges Devine, Glecier, Jenkins, LeFevour, McCollom, McNulty, Melograne, Murphy,Oakey, Reynolds, Sollie, and Scaccheti. See Pahis, supra note 31, at 1-6, 8.

56. Judges Bates, Brennan, Cain, Collins, Coruzzi, DeSaulnier, Harris, Hastings, Hogan,Jenkins, Murphy, Nixon, Reynolds, Shiomos, and Thoma, as well as Chief Judge Claiborne.Id. at 1-7.

s7. Judges Adams, Alonzo, Greer, Holzer, and Malkus. Id. at 1, 3, 6, 7.58. Judges Devine, Glecier, Holzer, James, LeFevour, McDonnell, Murphy, Olson, Reynolds,

Seaman, and Sodini. Id. at 1-6.59. Justice McCann. Id. at i.6o. Judge Salerno. Id. at 5. These numbers add up to forty-five, as some judges accepted bribes

in different types of cases.

61. Judge LeFevour was responsible for accepting over 2500 traffic bribes alone. Id. at 2.


handled by the courts (54%).62 Information on the number of instances inwhich bribes were transacted in criminal cases and civil cases is a bit clearer,but is still less than precise. For example, one judge was accused of requesting a"loan" from a lawyer when that lawyer had "a block" of cases before thejudge.6 3 Comparing the number of corrupt decisions in the two types of casesfor which there is specific information, it appears that discovered bribery incriminal cases holds a dominant position over discovered bribery in civil cases.The ratio of 70 corrupt criminal decisions to 38 corrupt civil decisionsrepresents a significant deviation from the 11:9 ratio of criminal to civil casesheard in state courts.14

B. Corrupt Actions: What the Bribes Bought

i. Criminal Cases

At least ten of the bribes paid in criminal cases sought to influence a judgeto directly reduce or suspend a sentence, probation, or fine, after conviction.6 s

Three bribes were paid to two judges so that they would influence the decisionof another judge to reduce prison sentences.66 One bribe was paid to a judge topersuade the prosecutor to drop charges against the defendant. 67 At least threebribes bought the dismissal of charges,68 while another bribe bought theordering of a new trial after the defendant was convicted. 69 In at least twocases, the briber sought or received a reduction in bail.7" In one case, a bribe

62. This is the percentage of incoming traffic cases in state courts in 20o6. COURT STATISTICSPROJECT, supra note 43, at 12. Removing drunk-driving cases from the traffic category toconform with the Court Statistics Project's definition of traffic cases would reduce thisoverrepresentation, but probably not significantly. See supra note 43.

63. Judge Holzer. See Pahis, supra note 31, at 3.

64. Criminal cases made up 21% of all incoming cases in state courts in 2006, while civil casesmade up 17% of the same. COURT STATISTICS PROJECT, supra note 43, at 12. Domestic andjuvenile cases made up 6% and 2%, respectively, of the 2006 incoming case load. Id. Addingdrunk-driving cases to conform with the Court Statistics Project's definition of criminalcases would only increase this overrepresentation. See supra note 43.

65. Judges Bates, Brennan, Collins, Coruzzi, Harris, Hastings, and Thoma. See Pahis, supra note31, at 1-3, 5-7.

66. Judges Brennan and DeSaulnier, as well as Chief Judge Claiborne. Id. at 1-3.

67. Judge Nixon. Id. at 4.

68. Judges Brennan, Cain, and Reynolds. Id. at 2-3, 5.

69. Judge Brennan. Id. at 2-3.

70. Judges Cain and Harris. Id. at 5.

1926

118:19oo 2009


was used to buy the quashing of subpoenas related to an investigation,7 whileanother bribe bought the acceptance of a habeas corpus petition.72 At least onebought the acquittal of the defendants.73

2. Civil Cases

Corrupt action varied in civil cases. Three judges were convicted of steeringcases toward settlements favorable to the bribing party, either through rulingon motions or advocating during settlement negotiations.74 Two judges offeredprivate consultations on cases to the lawyer before them.7

' Two judges handeddown favorable monetary judgments.76 In one case, the judge extorted moneyfrom a client of his former law firm by threatening to dismiss and impede theprocessing of their future cases. 77 In another set of cases, the judge wouldsolicit loans from lawyers trying their cases before him (which incidentallywere never paid back), although no explicit promises or threats were made. 78

3. Traffic Violations

Bribing parties in traffic violation cases generally sought the dismissal ofthe case or, in the case of drunk driving, sometimes the lenient sentence ofsupervision. In the case of Judge Melograne, violations were dismissed beforethe police officers arrived to testify or after they were ordered away from theproceedings. Judge Melograne also conspired with the supervising judge at theStatutory Appeals Division to influence other judges. In the case of many of theGreylord judges, police officers-who often acted as middlemen passing thebribes to the judges -would sometimes take a cut of the bribe to change theirtestimony in court to allow for a dismissal.79

71. Chief Judge Claiborne. Id. at 3.72. Judge Jenkins. Id. at 6. The fourteen bribes that Judge Harris accepted were paid for the

dismissal of charges, reductions in bail, and reductions in sentencing, in proportionsunknown to this author.

73- Judge Murphy. Id. at 2.

74. Judges Adams, Greer, and Malkus. Id. at 6-7.75- Judges Greer and Malkus. Id. at 6-7.76. Judges Malkus and Adams. Id. at 7.

77. Judge Alonzo. Id. at i.78. Judge Holzer. Id. at 3.79. See, e.g., Judges LeFevour and Murphy. Id. at 2.

1927


4. Administration of Cases

In one case, a lawyer bribed the supervising judge to assign cases to judgesof the lawyer's choosing. In another case, the judge set higher attorney's feesfor a court appointed lawyer in exchange for a kickback 8 ' In yet another case, ajudge repeatedly increased bail for defendants in exchange for a kickback fromthe bail bond agency.8s Many of the Greylord judges had an elaborate process,by which lawyers would bribe judges for the opportunity to solicitunrepresented defendants within the court, court clerks would steer thesedefendants to the paying lawyers, and the judges would share a cut of the feesthat they would assign the lawyers for their work.s3

C. Prices: How Much Was Paid

The bribes studied in these cases varied greatly in value, from a bag of icefor the dismissal of a traffic violation8 4 to a $15o,ooo payment for the reversalof a forfeiture order of $845,000 in a criminal case. 8

' Bribes in criminal caseswith potential jail time ranged from $1oo for the favorable treatment of auto-theft defendants 6 to $1oo,ooo (half of which went to the middleman) for atleast a forty-two-month reduction in jail time.17 In the civil cases studied,bribes often were made with gifts and services or loans that were not or wereonly partially repaid. The sample size of civil cases is even smaller, and evenharder to evaluate, as three of the five corrupt judges served on the same courtand dealt with the same briber on multiple occasions. Nonetheless, it appearsthat very valuable judgments or settlements yielded substantial yet comparablysmall bribes. For example, $7 million worth of corrupt settlements yielded a$20,500 reward for one judge.88 The disposal of traffic citations and finesyielded understandably smaller bribes for judges. There are records of $40 to$1oo bribes and of gifts ranging from guns to jewelry to a bag of ice and a

So. Judge Greer. Id. at 6.81. Judge James. Id. at 1.82. Justice McCann. Id. at 1.83. Judges Devine, Glecier, Holzer, LeFevour, McDonnell, Murphy, Olson, Reynolds, Seaman,

and Sodini. Id. at 2-6.

84. Judge Jenkins. Id. at 6.

85. Judge Hastings. Id. at 6.

86. Judge Hogan. Id. at 5.

87. Judge Collins. Id. at 6.

88. Judge Adams. Id. at 7.

1928

118:19oo 2009


pound of ham.8 9 Small gifts do have the potential to add up over time,however, evidenced by the Greylord judge charged with accumulating over$10o,ooo from ticket fixing over a ten-year period of time.9" Judges involvedin kickback schemes took between io% and 50% of the attorneys fees or bail-bond premiums that they were responsible for assigning.

D. Risk of Detection and Tip-Offs

The evidence regarding our ability to detect judicial corruption is generallynot very encouraging. At least 29 of the 38 judges had engaged in previouscorrupt acts prior to being caught.91

i. Prosecutorial Leverage

The data suggest that the most successful method for discoveringcorruption is the employment of prosecutorial leverage by the DOJ. Of the 20

investigations that ensnared the 38 judges in the sample, 1o (55%) wereinitiated through information obtained through prosecutorial leverage.92

Criminal defendants were responsible for outing the largest number of corruptjudges, in exchange for more lenient sentences. In 6 investigations (30% oftotal investigations and 50% of criminal), the defendant or his agentvoluntarily contacted the authorities after establishing a corrupt relationshipwith the judge. 93 In at least two cases, the briber went to the authorities afterhe received a heavier sentence than he had bargained for.9 4 This would seem toconfirm the hypothesis that dealing corruptly with criminal defendants, towhom this leverage can be extremely valuable, represents a significant risk tojudges.

89. See, e.g., Judges Cain, Jenkins, LeFevour, Murphy, Oakey, and Salerno. Id. at 2, 5-6.

go. Judge LeFevour. Id. at 2.

gi. Judges Adams, Alonzo, Brennan, Cain, Coruzzi, Devine, Glecier, Greer, Harris (fourteencases through two lawyers), Hogan, Holzer, James, Jenkins, LeFevour (thousands of cases),Malkus, McCollom, McDonnell, McNulty, Melograne (hundreds of cases), Murphy, Oakey,Olson, Reynolds, Salerno, Scacchetti, Seaman, and Sodini, as well as Justice McCann(thirty-seven payments from bondsman) and Chief Judge Claiborne. Id. at 1-8.

92. Judges Bates, Brennan, Cain/Harris/Shiomos, Collins, Coruzzi, Hastings, Jenkins,Scacchetti, and Thoma, as well as Chief Judge Claiborne. Id. at 1-7.

93. Judges Brennan, Collins, Coruzzi, Scacchetti, and Thoma, as well as Chief Judge Claiborne.Id. at 1-3, 6-7.

94- Judge Thoma and Chief Judge Claiborne. Id. at 3-7.

1929


Prosecutorial leverage was also critical in the initiation of three otherinvestigations. In the cases of Judges Hastings and Jenkins, the judges'middlemen became FBI informants in return for leniency with respect tounrelated charges. In another investigation, the briber became a governmentinformant after an unrelated investigation uncovered other corrupt dealings.9

While it was not instrumental in initiating the investigation, prosecutorialleverage was vital to the success of Operation Greylord, in which multiplejudges, lawyers, and clerks became government witnesses in return. forleniency. 96

2. Judicial Conduct Organizations and Uninterested Tip-Offs

Of the twenty investigations studied, it appears only two were initiated byJCOs97 and only the investigation of Judge Sollie was handled exclusively bythe JCO; the other JCO investigation led to criminal charges. OperationGreylord began when a disillusioned judge approached the FBI with noexpectation of personal gain. 8 That it too over ten years after the expansivering of corruption began for someone to come forward provides a soberingwarning of the hazards of relying too heavily upon voluntary and disinterestedtip-offs to fight corruption. Indeed, the lead prosecuting attorney forOperation Greylord lamented that he was not aware of a single lawyer whocame forward voluntarily to complain about what was widespreadcorruption. 99 Underscoring this point is the fact that no investigation ofjudicial bribery in this sample was initiated by Congress, a legislature, or by ajudicial council.

3. Judicial Extortion

In two cases in which the judge initiated the corrupt relationship and triedto extort money from attorneys, the attorneys themselves recorded theconversations and went to the authorities,"' suggesting a potential danger in

95. Judges Cain, Harris, and Shiomos. Id. at 3-5.

96. For a first-hand account of the investigation, see BROCTON LocKWOOD WITH HARLAN

MENDENHALL, OPERATION GREYLORD: BROCTON LocKWOOD'S STORY (1989).

97. Judges Adams/Greer/Malkus and Sollie. See Pahis, supra note 31, at 1, 6-7.

98. See generally LOCKWOOD, supra note 96.

99. Maurice Possley, Lauryers" 'Code of Silence' on Greylord Assailed, CHI. TRIB., Nov. 23, 1985, § 1,at 1.

ioo. Judges Alonzo and James. See Pahis, supra note 31, at I.

1930

118:19oo 2009


judge-initiated extortion. The case of Judge Holzer, who extorted money fromlawyers with business in front of the court on at least three occasions, however,signals that lawyers may be reluctant to report judges for fear or retribution orfor hopes of future payoffs.

4. Multijudge Corruption Rings

The incidence of multijudge corruption schemes is also noteworthy. Therecould be reasons to believe that multijudge involvement stymies investigationsand reduces risk. Multiple players, however, also may serve to increase the riskof defection. First, the more judges involved, the greater the chance that anhonest judge or a private party will take notice, prompting an investigation.Second, once an investigation begins, each corrupt judge will be caught in aclassic prisoner's dilemma.' The more judges involved, the greater theincentive that a judge faces to defect. The evidence from this sample bears thisargument out. In all three of the multiparty corruption schemes, corrupt judgesbecame witnesses for the prosecution in exchange for leniency.

5. First-Mover Risk

Section II.B hypothesized that the first-mover risk might lead to long-termcorrupt relationships between parties (or lawyers) and judge's. The sampleneither supports nor contradicts this hypothesis. All of the fifteen Greylordjudges engaged in multiple acts of corruption before being caught, and at leastthirteen of them accepted bribes from multiple parties or lawyers.' °2 Of thenon-Greylord judges, at least fifteen of the twenty-two engaged in multiplecorrupt acts, with at least eight0 3 dealing with multiple parties or lawyers andat least seven 0 4 dealing multiple times with the same lawyer or party. Only twojudges were removed because an attorney reported their offer.' 5 But given the

lo. For an introduction to the prisoner's dilemma, see ANATOL RAPOPORT & ALBERT M.CHAMMAH, PRISONER's DILEMMA (1965).

102. This is true for all except Judges McDonnell and McNulty, for whom the evidence is notclear either way. See Pahis, supra note 31, at 4, 6.

103. Judges Brennan, Coruzzi, Harris, Jenkins, Melograne, Shiomos, Sollie, and Thoma. Id. at1-8.

104. Judges Adams, Cain, Greer, James, Malkus, and Scacchetti, as well as Justice McCann. Id. at1, 5-7.

1o5. This might suggest that our assumption was correct that judges face higher first-mover riskand therefore abstain from making the first move, leaving the parties and lawyers to bear the


number of bribes that changed hands before the judges were caught and giventhe number of parties involved, it would seem that any explanation would needto allow that corrupt judges, parties, and lawyers find ways to mitigate thefirst-mover risk and signal their openness to making a deal without leavingthemselves vulnerable to reporting.

The use of middlemen was one method that some of the judges used toaccomplish this. While at least fourteen judges dealt directly with the party,1o6at least fifteen judges dealt through middlemen, including friends of the judgeor of the party," 7 bail bondsmen,1o8 bailiffs,' 9 third-party attorneys,"0

policemen,"' clerks," 2 and unidentified middlemen."3 At least six judges dealtdirectly with lawyers who appeared before them." 4

V. ANALYSIS

The most striking findings of this study are the disproportionately highnumber of uncovered bribes related to traffic tickets vis-A-vis bribes made incriminal and civil cases, and the disproportionately high number of bribes incriminal cases as compared to bribes in civil cases. While it is unclear by howmuch traffic bribes are overrepresented in the sample, it is clear theoverrepresentation is significant: Thousands of bribes were exchanged intraffic cases compared to around one hundred bribes made in all other cases.The overrepresentation of criminal cases vis-A-vis civil cases is also significant.This begs the question of whether these patterns reflect the incidence ofjudicial corruption in reality, or whether they represent a difference in relativerates of detection. Let us first examine the discrepancy between bribery in

risk. The dataset cannot confirm this, however, as it does not include criminal or civil partieswho were convicted for offering a rejected bribe.

1o6. Judges Alonzo, Hogan, Holzer, James, Jenkins, Murphy, Nixon, Oakey, Reynolds, Salerno,Sodini, and Sollie, as well as Justice McCann and Chief Judge Claiborne. See Pahis, supranote 31, at 1-6.

107. Judges Bates, Brennan, and Scacchetti. Id. at 1-3.

1o8. Judge DeSaulnier. Id. at i.

1o9. Judge Sodini. Id. at 4.

11o. Judges Coruzzi and Hastings. Id. at 2, 6.

iii. Judges Hogan, LeFevour, McCollom, and Murphy. Id. at 2, 4-5.

112. Judge Reynolds. Id. at 3.

113. Judges Collins, Salerno, and Thoma. Id. at 5-7.

114. Judges Adams, Cain, Greer, Harris, Malkus, and Shiomos. Id. at 3-7.

118:19oo 2009


criminal versus civil cases before moving on to addressing the frequency ofbribery in traffic cases.

A. Criminal Versus Civil Cases

This Section argues that bribery in criminal cases is likely to presentrelatively more serious risks to the judge involved. This suggests that thesample overrepresents the amount of corruption in criminal cases, whileunderrepresenting the amount of corruption in civil cases. ' Determining thelevel of bias in the sample requires evaluating the risk differential and otherfactors that drive the supply and demand of these respective corrupt decisions.

1. Risk of Detection

As discussed in the previous Part, six of the twelve investigations thatensnared the sixteen judges caught in bribery schemes related to criminal caseswere initiated by the criminal defendant. The defendant became a governmentinformant in exchange for a lenient sentence in the same case before the bribedjudge. By making the corrupt arrangements and then notifying the authorities,the defendants were able to save money while still achieving a comparableresult. The plea-bargaining authority of prosecutors appears then to create astrong incentive for the criminal defendant to defect in a bribery transaction.This raises the risk of detection for the judge accepting the bribe by addingwhat is akin to a first-mover risk to all such transactions, whether they are thefirst bribes exchanged between the two parties or not. Just as the moral gainsof reporting induces offerees to reject and report bribes that otherwise wouldbe advantageous to both parties, the plea-bargaining power of prosecutorsinduces some defendants to do the same. The judge, therefore, mustcontemplate the net benefits the briber could reap from reporting the bribe,even in the case where the defendant makes the first move. When the judgebelieves that the defendant could get a better deal from prosecuting authorities,the judge should refuse to engage. While in theory, prosecutors should alwaysbe able to provide a better deal by reducing a defendant's sentence for free, inpractice this may not always be the case. There are risks that authorities maynot be able to prosecute the judge, that prosecutors may not be open to

115. The bias toward large-scale or newsworthy corruption, discussed in Section III.D, wouldnot seem to skew the sample toward criminal cases, as none of the multijudge corruptioncases were related to criminal trials. If anything, this sample bias would lead to anoverrepresentation of civil case bribery, since, as was already noted, three of the four judgesinvolved in civil case bribery served and acted corruptly together.


eliminating the defendant's charges or sentence, or that they may choose to

prosecute the defendant for attempted bribery. It is difficult to know how largethese risks are-the fact that the sample shows that some prosecutors are

willing to make deals with defendants who initiated the bribe themselves may

signal that the risks are not that large-but they most likely exist and createenough uncertainty that the expected value provided by the prosecuting

authority is not always greater than the value provided by a corrupt judge.

Because judges face the risk that the defendant will choose to defect, thenatural inclination will be to demand a larger bribe to compensate for that risk.But just as in our first-mover model, increasing the bribe demanded decreases

the gain on the part of the defendant, leading to a greater risk of defection.Graphically we can show this as follows. While in a civil case, a judge would

attempt to bargain for a bribe as close as possible to the maximum that thelitigant would be willing to pay (Bp), here a judge would not accept any offer

that would provide the briber less value than he would receive by reporting the

bribe, which is represented as PLEA, for please bargain. Here, the judge would

only accept a bribe that was less than (Bp-PLEA).

Figure S

Judge's Willingness To Sell

Party's Willingness To Pay I

0 gBi.) Bft.) - PLEA BR-)

S BRIBE

In Figure 5, there is- room for a successful bribe as the judge is willing toaccept a bribe (B) that is less than (Bp - PLEA). In general, however, we wouldexpect the possibility of defection to reduce the number of bribes exchanged

between judges and criminal defendants. We can show this through the

following example. First, assume that there is a uniform distribution of judges'willingness to accept minimum bribes and that this minimum bribe threshold

is distributed between bribes that are $1ooo and bribes that are $4000. Next,assume that criminal defendants are willing to pay up to a certain bribe inbetween that range, say $3000. Absent the risk of defection, bribes will be

transacted in two-thirds of all cases (in all cases in which judges are willing toaccept as little as a $300o bribe, represented by both shaded regions). But after

accounting for the prosecutor's plea-bargaining powers, which here we assumeto offer a value of $iooo, we see that bribes will only be transacted in one-third

of criminal cases (represented by the lightly shaded region). This analysis

suggests that, all else equal, we should expect less bribery in criminal casesthan in civil cases, in which there is no analogous incentive for defection.

. 118:19oo 2009


Figure 6

0 12 3 4

S Bribe (in thousands)

If the model is correct in its prediction that judges will shy away fromaccepting bribes within the zone of defection, why does the sample suggest theopposite? Indeed, six of the sixteen judges removed or convicted for acceptingbribes in criminal cases were caught precisely because the defendant defected.To answer this question, we first must remind ourselves that we are lookingnot at a sample that is representative of bribery as it exists, but of bribery thatis discovered. This sample in all likelihood overrepresents cases of high-riskbribery. Still, the model predicts that judges will rationally avoid accepting anybribes that fall within the zone of defection. Why, then, are any such bribesfound in our sample? The fact that they are found suggests that the judgescaught in this way either miscalculated or failed to perceive the risk. ,6 Equallyplausible is the possibility that the defendants themselves miscalculated the

116. Many studies in behavioral law and economics have questioned the individual capacity forrational decisionmaking, even when the decisionmaker possesses complete information. SeeRobert A. Prentice, The Case of the Irrational Auditor: Behavioral Insight into Securities FraudLitigation, 95 Nw. U. L. REv. 133 (2000) (providing a brief overview of the literature).Individuals tend to be overly optimistic about their chances of success and think that theycan beat the odds. See, e.g., Donald C. Langevoort, Organized Illusions: A Behavioral Theory ofWhy Corporations Mislead Stock Market Investors (and Cause Other Social Harms), inBEHAVIORAL LAw AND ECONOMICS 144, 149 (Cass R. Sunstein ed., 2000) ("One of the mostrobust findings in the literature of individual decision making is that of the systematictendency of many people to overrate their own abilities, contributions, and talents. Thisegocentric bias readily takes the form of excessive optimism and overconfidence, coupledwith an inflated sense of ability to control events and risk."); Cass R. Sunstein, Introductionto BEHAvIORAL LAw AND ECONOMICS, supra, at 1, 4 ("Even factually informed people tend tothink that risks are less likely to materialize for themselves than for others. Thus there issystematic overconfidence in risk judgments .. ").

1935


benefit provided by the prosecuting authority. Given the potential uncertaintysurrounding the gains achievable through defection, this is understandable.Unless the DOJ and state prosecutors develop a transparent policy of offeringimmunity to all defendants who can deliver to them a corrupt judge, 17 neitherthe judge nor the defendant will be aware of what can be gained throughreporting the bribe. So while in theory there may be a clearly delineated zone ofdefection in which all bribes will be reported, in practice the edges of this zonemay be rather fuzzy, creating the risk that even well-calculated bribes acceptedon the part of the judge will be reported. Unlike other risks, this one cannot becompensated by a larger bribe and so it should lead to even fewer corruptexchanges in criminal cases. Those judges who chose to brave the risk weremore likely to appear in the sample.

The risk to judges accepting bribes in criminal cases is even greater whenone considers that the government, being the prosecuting party, is likely tohave a greater interest in corruption in these cases and, having front-row seatsto the trial, is more likely to be attuned to suspect behavior by judges.Moreover, once an investigation into bribery in a criminal case is launched, itfaces a higher probability of success. The prosecution is more likely capable ofinducing defection even in cases where the briber would have preferred not todefect ex ante. Unlike investigations into civil corruption, the prosecution willbe able to offer more than just immunity in return for incriminatinginformation against the judge.

Because there seems to be no reason to believe that a judge would have agreater incentive to report a bribe in a civil case than in a criminal case, it isreasonable to conclude that corruption is less likely to be discovered in civilcases than in criminal cases, and that all else equal, judges are less likely toaccept bribes in criminal cases. This conclusion is significant and provides atleast one reason for believing that criminal cases are overrepresented and civilcases are underrepresented in our sample.

2. Actual Incidence

Whether, in fact, we expect there to be more bribery in civil cases than incriminal cases depends upon whether the willingness to pay of civil litigants isgreater than that of criminal defendants. It also depends upon whether weexpect judges, the risk of defection and the greater risks of detection in general

117. The use of informants and the prosecutorial "deals" given to them is one of the most opaquepractices in the judicial system and suffers from great inconsistencies in application.Alexandra Natapoff, Snitching: The Institutional and Communal Consequences, 73 U. CIN. L.REV. 645, 648, 654 (2004).

1936

118:19goo 2009


notwithstanding, to be more or less willing to accept bribes in criminal or civilcases. To address what our sample and model can tell us about what types ofcorruption go undetected, this Subsection explores both of these questions inturn.

a. Demand

There are at least five reasons for believing that demand for corruptionmight be relatively greater in criminal cases than in civil cases and at least onereason for believing that the opposite is true. First, criminal defendants mayface greater stakes; all else equal, a criminal defendant may value freedom fromthe average jail sentence more than a civil party values the average monetaryaward from litigation. Given the great variance in value of the stakes in civiland criminal claims, however, it is unclear how much explanatory power thishypothesis provides. A second reason for why there might be relatively moredemand for corruption in criminal cases is that criminal defendants facerelatively lower costs of detection. While a criminal defendant might face agreater risk of detection, given the government's direct interest in the case, thepotential costs of detection in terms of reputation and employmentopportunity costs are probably relatively low, as they may already faceincarceration. The willingness of the government to grant immunity to thedefendant for turning state's evidence further reduces the costs of detection.Civil litigants, on the other hand, especially those engaged in high-stakeslitigation, might have greater personal, professional, or corporate reputationsto uphold. Third, even holding the stakes as well as the risk and cost ofdetection equal, the criminal defendant should be willing to pay more for thesame value, as he has a sort of insurance policy against losing the value of thebribe. Should the judge not deliver the results bargained for, the criminaldefendant has the ability to defect to the authorities and achieve a comparableresult. The civil party has no such option reasonably available. Fourth, criminaldefendants may face lower morality costs and be less risk averse. If they didindeed commit the crime they are accused of, their status as a defendant couldbe a signal of moral flexibility that translates, according to our model, into awillingness to offer larger bribes.

A fifth reason that criminal defendants may have a greater demand forcorruption is that the decisions their bribes buy are probably less likely to beappealed and overturned." 8 While the government can appeal a range of

ii8. Vikramaditya S. Khanna, Double jeopardy's Asymmetric Appeal Rights: What Purpose Do TheyServe?, 82 B.U. L. REv. 341, 350 (2002).


judicial decisions -including orders. of dismissal before trials begin andjudgments notwithstanding the verdict after trials end -prosecutorial appealsof midtrial dismissals, orders to suppress evidence, and decisions that leaddirectly to an acquittal are forbidden by the Fifth Amendment Double JeopardyClause.' 9 This leaves corrupt judges with discretion to carry out the briber'swish in a manner that is nonreviewable.'2 The judges in this sample appearedto prefer reducing sentences, an action also relatively protected from appellatereview. While prosecutorial sentence appeals were held constitutional by theU.S. Supreme Court in 198o,2 their use has been restricted both at the stateand federal levels.' Federal law restricts federal prosecutorial appeal tosentences allegedly "imposed in violation of the law,"'2 3 "imposed as a result ofan incorrect application of the sentencing guidelines,"' 4 "less than the sentencespecified in the applicable guideline range,"'25 or "imposed for an offense forwhich there is no sentencing guideline and [are] plainly unreasonable. '

,,,6

Meanwhile, relatively few states have adopted statutes that enable prosecutorialsentence appeals at all, 2 ' and many of the states that allow such appeals restrictthem to sentences that depart from sentencing statutes."' Relieving some of

119. U.S. CONST. amend. V. ("No person shall ... be subject for the same offence to be twice putin jeopardy of life or limb ...."). The Fourteenth Amendment applies the prohibition ondouble jeopardy to the states. Benton v. Maryland, 395 U.S. 784, 794 (1969); see AnneBowen Poulin, Government Appeals in Criminal Cases: The Myth of Asymmetry, 77 U. CiN. L.REV. 1, 4, 51-52 (20o8) ("Mid-trial rulings that fold into the ultimate verdict are insulatedfrom government requested review as well. In addition, substantive issues arising inconnection with jury instructions and pro-defendant evidence rulings are frequently beyondthe reach of government appeal."). Acquittals purchased through bribes, however, are notsubject to the prohibition of double jeopardy. United States ex rel. Aleman v. Circuit Courtof Cook County, 967 F. Supp. 1022 (N.D. I11. 1997), affid, 138 F.3 d 302 (7th Cir. 1998).

120. See Poulin, supra note 119, at 52 ("A trial court is sometimes able-intentionally or not-tostructure its rulings to preclude appellate review.").

121. United States v. DiFrancesco, 449 U.S. 117 (198o).

122. See Christina N. Davilas, Note, Prosecutorial Sentence Appeals: Reviving the Forgotten Doctrinein State Law as an Alternative to Mandatory Sentencing Laws, 87 CORNELL L. REV. 1259, 1266(2002).

123. 18 U.S.C. § 37 4 2(b)(1) (2000).

124. Id. § 3 7 4 2(b)(2).125. Id. § 3742(b)(3).

126. Id. § 3742(b)(4).

127. See Davilas, supra note 122, at 1266.

128. Id.; see, e.g., ARiz. REv. STAT. ANN. § 13-4032(5) (2001); FLA. STAT. ANN. § 924.07(1)(e)

(West 2001); KAN. STAT. ANN. § 21- 4 7 21(d), (e) (2007); LA. CODE CRIM. PROC. ANN. art.882 (2008).

1938

118:19oo 2009


the restrictions on sentencing appeals would decrease the value of corruptionwhile avoiding the more restrictive approach of minimum sentencing laws.

There is at least one reason, however, for believing that that there may begreater demand for corruption in civil cases: the contingency payment method.As shown in the sample, attorneys, who deal more closely and repeatedly withjudges than the parties, seem to be an effective conduit for corruption in bothcivil and criminal cases. In the civil cases involving Adams, Greer, and Malkus,for example, the corrupt attorney developed ongoing relationships with thejudges and conspired with them to set attorneys fees in settlements. Thegreater use of the contingency payment method in civil cases may create agreater incentive for civil lawyers to bribe judges. 129

b. Supply

While there are persuasive reasons for believing that demand forcorruption in criminal cases might be relatively greater, there are at least fivepersuasive reasons for believing the overall supply of corruption in criminalcases is relatively lower. The first four have been discussed extensively in theabove Subsection and will not be elaborated upon again here, though theydeserve brief mention. First, the risk of defection that judges face will causethem to refrain from accepting bribes that, in other circumstances, wouldbenefit both parties. Second, the uncertainty surrounding the value to thebriber of reporting the bribe creates additional risk for bribes that theoreticallyfall outside the zone of defection. Third, the government has a greater interestin corruption in cases in which it is a party and is in a better position to detectsuspicious behavior on the part of the judge. Fourth, investigations intobribery in criminal cases are more likely to be successful even in the absence ofdefection, as the government can offer incentives for the defendant to defectthat go beyond immunity to charges of bribery. All of these factors point togreater risk for judges who accept and solicit bribes in criminal cases, which, allelse equal, should reduce the incidence of such corruption.

The last factor that may reduce the willingness or ability of judges to sellcorrupt criminal decisions is the prevalence of minimum sentencing laws.13

129. Using the contingency fee method in criminal cases has generally been found to be unethicaland prohibited. See Adam Silberlight, Gambling with Ethics and Constitutional Rights: A Lookat Issues Involved with Contingent Fee Arrangements in Criminal Defense Practice, 27 SEATTLE U.L. REv. 805, 805 (2004).

130. See 24 C.J.S. Criminal Law § 2005 (2006).


These laws restrict judges' discretion with regards to what the sample reflectsis the most common corrupt action taken in criminal cases.13 '

There is at least one factor, however, that may increase the willingness ofjudges to make the first move and solicit a bribe in a criminal case. A criminaldefendant's position is most likely a sign of moral flexibility. The moral gainthat criminal defendants receive from reporting a bribe offer therefore might berelatively lower than the moral gains accruing to civil parties. It would notseem that any of the other factors that influence supply of corruption in ourmodel would lead intuitively to greater or lesser supply of corruption incriminal cases. For example, it does not appear that judges handling criminalcases would face greater detection costs than judges handling civil cases. Theprevalence of courts that handle both criminal and civil cases makes this aneffectively moot point.'32

c. Uncertain Conclusions

The model leads us to believe that while the demand for corruption incriminal cases is greater than in civil cases, the supply of corruption is mostlikely lower. It remains unclear, therefore, whether we should expect a greateror lesser incidence of bribery in civil cases vis-A-vis criminal cases. This findingnotwithstanding, the theory of differential rates of detection continues to be arelevant and explanatory piece of the puzzle, leading to the conclusion thatbribery in civil cases is less likely to go detected.

3. Bias and Alternative Explanations

The bias toward large-scale or newsworthy corruption, discussed in SectionIII.B, would not seem to help explain the sample's disproportionately lownumber of bribes in civil cases relative to the number of actual civil cases.While Operation Greylord did net an additional three judges who accepted

131. See supra Subsection IV.B.i. Since most of the corrupt acts in criminal cases went unnoticeduntil a party to the bribery notified the authorities, and because the investigations andprosecutions relied on other key testimony and evidence, there should not be a significantbias in sample from dealing only with discovered corruption. We might infer, therefore, thatsentencing reduction is the means preferred by judges for achieving the desired result ofcriminal defendants. Other options for corrupting criminal cases include dismissing thecase, issuing a judgment notwithstanding the verdict, and acquitting the defendant in abench trial-all of which are likely to call greater attention to the judge and allow forappellate review.

132. COURT STATISTICS PROJECT, supra note 43, at 13.

1940

118:1900o 2009


bribes in criminal cases and one judge who accepted bribes in civil cases, thenumber of criminal bribes continues to far exceed the number of civil bribeseven without accounting for these judges. Moreover, of those judges caughtoutside the Greylord net, three of the four judges involved in civil case briberyserved and acted corruptly together -the same number of judges who servedtogether and accepted bribes in criminal cases.

Another possible explanation for our sample's disproportionately lownumber of bribes in civil cases is that they are more likely to be handled byjudicial conduct organizations or judicial councils. Given that the discovery ofbribery in criminal cases is most often the result of a defendant seekingleniency, it seems reasonable to believe that the briber would notify law-enforcement authorities who could affect his sentence. The corollary is thatcivil cases are relatively more likely to be reported to judicial conductorganizations or judicial councils. Because JCOs and judicial councils are notendowed with prosecutorial leverage and are likely to only come across the caseex post, they may not be able to prove there was consideration in a transactionbetween a judge and a litigant or lawyer. Some quid pro quo transactions,then, would likely be classified as gifts, rather than as bribes.

It is unclear, however, how much this reclassification of bribes as giftswould bias the sample. After all, if a judge is removed by a judicial conductorganization, the case becomes public, opening the door for a follow-up DOJprosecution. The ex post nature of the investigation (as opposed to criminalinvestigations which can make use of informants to catch the judge in the act)might hinder prosecution, as would the absence of any meaningfulprosecutorial leverage. Bribing parties in civil cases, of course, do potentiallyface criminal charges for corruption, giving them an incentive to testify againstthe judge in exchange for immunity. But, unless the civil party is caught in theact of bribing or is implicated by a middleman, this incentive is limited. Thejudge, after all, would face little incentive to admit to wrongdoing unless hewere part of a multijudge bribery scheme and engulfed in the type of prisoner'sdilemma described above. Therefore, to the extent these reclassifications biasthe sample, they also confirm the conclusion of this study that bribery in civilcases is less likely to be discovered.

The bias might be exacerbated when taking into account resignations. If ajudge resigns amid an investigation, and the case against him is closed, it isunclear whether the JCO or judicial council is required to refer the case to theDOJ or does so in practice. Given concerns of maintaining judicialindependence and integrity, there are reasons to believe JCOs and judicialcouncils would be reluctant to do so. Indeed, the promise to close the casecould be a valuable bargaining chip for encouraging a resignation.

To the extent the judges engaging in these acts are removed or leave thejudiciary, the accountability system would seem to work effectively. There are


at least two reasons, however, to believe that civil corruption is still relativelyunderdetected. First, these institutions are comparatively limited in their abilityto discover and deter corruption. Without prosecutorial leverage, discoveringand investigating civil bribery, even as gifts, will still be comparatively moredifficult to do than discovering and investigating bribery in criminal cases.Second, while disciplinary sanctions and removal from office can be strongdeterrents, they are less severe than the threat of criminal prosecution.Combining these factors leads us to conclude that bribery in civil cases presentslower costs to both the judge and the briber, possibly giving rise to a higherrelative incidence rate than the sample shows.

B. Traffic Bribes

Bribery in traffic court would seem to fit in the same model as bribery incriminal court. After all, a defendant in traffic court presumably could tradeinformation on a corrupt judge for the dismissal of his ticket. It presumablywould be easier for the authorities to waive a traffic fine than a serious criminalcharge for incriminating information on a judge. And yet the sample does notshow any instances in which a traffic defendant voluntarily defected to theauthorities. 3 ' This is not necessarily surprising when one considers thatdefection is not costless. Successful defection could require time-intensivecooperation with the authorities. It also carries a risk of failure and could bringpossible charges against the defendant. Given the small stakes in traffic cases,these costs and potential risks are probably rarely worth incurring.

This might help explain why the sample shows that those traffic judgeswho accepted bribes did so repeatedly and extensively before being caught.Operation Greylord unveiled over ten years worth of ticket fixing whichallowed one of the fifteen judges to amass over $1oo,ooo. While repeat briberymay increase the rate of detection if it continuously raises red flags, as with thecase of Judge Melograne, the alarmingly high bribe-to-judge ratio (in thehundreds) suggests that the risk involved with corruption in traffic courts isrelatively small and that the judges involved were able to successfully mitigatethe first-mover risk. It is possible, of course, that Operation Greylord, whichensnared fifteen judges, is an outlier. Even after removing the fifteen Greylordjudges from the study, ticket fixing remains overrepresented relative to the

1942

133. Judge Scacchetti was caught when the employer of one of his traffic defendants was arrestedon unrelated charges and told the authorities about their prior corrupt dealings. See Pahis,supra note 31, at i.

118:i9oo 2009


number of traffic cases handled, as Judge Melograne alone was responsible forhundreds of corrupt transactions.

There is no strong rationale for removing Judge Melograne from the study,as there are at least three additional strong reasons for believing that theabundance of traffic-related bribes in the sample is an accurate reflection ofreality. First, given the small stakes involved, the interest that the state has inseeing a particular case through is probably weak. Second, the relatively littleevidence involved in traffic hearings leaves the judge in a position to exercise awide amount of discretion that can be abused at a low risk to the judge and thebriber. Finally, traffic court judges do not possess the same prestige or salary ofjudges higher in the judicial hierarchy, leading them to face lower detectioncosts. Inputting all of these factors into the model in Part I suggests thewillingness of judges to supply corrupt decisions in traffic court is most likelyquite high. This, along with the low stakes involved, explains the low value ofthe bribes changing hands.

CONCLUSION

While the small sample size of corrupt judges limits the certainty of ourfindings, the study suggests there is a troubling gap in our efforts to preventand prosecute judicial corruption. Of the thirty-eight judges studied in thiscase, thirty had engaged in corrupt acts other than the ones' that led directly totheir removal or conviction. That they were eventually caught is heartening,but it remains unclear how many other cases are being overlooked.

Even assuming these judges comprise a large share of a very small group of"bad apples," the many instances in which they were able to act corruptlywithout consequence is revelatory of -deficiencies in our anticorruptioninstitutions. This Note has attempted to shed light on these deficiencies byinvestigating the incentives that drive judicial bribery. Both the model and thesample suggest that corruption in civil and traffic cases seems especially proneto going undetected as compared to bribery in criminal cases. The discovery ofbribery in the latter type of case is largely due to the incentive that criminaldefendants have to report the corrupt judge in return for lesser charges in theirpresent proceeding. This same incentive, however, is not found when bribesare made by lawyers, civil parties, or traffic defendants. Without it, unearthingcorrupt relationships depends upon rare tips by third parties.

Expanding upon the sample of judicial bribes analyzed in this Note wouldprove helpful in establishing the robustness of these findings. Ways to bolsterefforts to prevent corruption in civil and traffic cases, including through thecreation of analogous incentives for defection, should be explored.

Corruption in Our Courts: What It Looks Like and Where It ...

Documents

Transcript of Corruption in Our Courts: What It Looks Like and Where It ...